Influence of Hydrogen on the Damage Behavior of IMC Particles in Al–Zn–Mg–Cu Alloys

Abstract

In recent years, it has been reported that intermetallic compound particles can suppress hydrogen embrittlement by hydrogen trapping. Some intermetallic particles in aluminum alloys, such as Al₇Cu₂Fe, have internal hydrogen trap sites; it is proposed that hydrogen embrittlement can be suppressed by preferential hydrogen partitioning in these sites. However, intermetallic compound particles act as fracture origin sites, and excessive addition degrades the mechanical properties of the material. In this study, we quantitatively evaluated the damage and decohesion behavior of intermetallic compound particles in high-hydrogen content 7XXX aluminum alloys by using in situ synchrotron radiation X-ray tomography. The results revealed that the hydrogen particles induced early high-strain localization, and the Al₇Cu₂Fe particles were damaged in that region due to its brittleness, resulting in early fracture. Hydrogen had no effects on the fracture and debonding behaviors of intermetallic compound particles, suggesting that the observed particle brittle fracture is dependent on their mechanical properties.

 

This Paper was Originally Published in Japanese in J. JILM 72 (2022) 411–419. Abstract and the captions of Figs. 1, 2, 4, 5, 6, 7, 9 and 10 are slightly modified.

1. Introduction

The strongest wrought aluminum alloys are Al–Zn–Mg–Cu alloys,¹⁾ which are commonly used in aircraft and other transportation equipment. By adding more Zn and/or Mg, this alloy is strengthened.²⁾ However, in the presence of hydrogen, the mechanical properties drastically deteriorate, and the susceptibility to hydrogen embrittlement increases.^3,4)

Various mechanisms have been proposed for the hydrogen embrittlement behavior of metallic materials. For example, Birnbaum et al. proposed hydrogen-enhanced localized plasticity (HELP),⁵⁾ in which the interaction between dislocations and hydrogen particles contributes to strain localization. Troiano et al. proposed hydrogen-enhanced decohesion (HEDE),⁶⁾ in which hydrogen reduces the bonding between metal atoms and induces cracking. Nagumo et al. proposed hydrogen-enhanced strain-induced vacancy (HESIV),⁷⁾ in which hydrogen promotes vacancy formation and growth during plastic deformation. In these hydrogen embrittlement models, the interactions of hydrogen with dislocations, vacancies, and grain boundaries are considered to be the dominant factors contributing to hydrogen embrittlement. However, the hydrogen binding energies of screw dislocations, edge dislocations, and vacancies in aluminum are as small as 0.08 eV/atom,⁸⁾ 0.17 eV/atom,⁸⁾ and 0.29 eV/atom,⁹⁾ respectively. Even for Σ5(012) high-energy grain boundaries, their binding energies are as small as 0.27 eV/atom,¹⁰⁾ creating uncertainty as to whether proposed hydrogen embrittlement mechanisms can be applied to aluminum.^11,12)

Recently, Tsuru et al. reported a mechanism whereby hydrogen accumulates at the MgZn₂ precipitate interface with high hydrogen binding energy, resulting in a decrease in the cohesive energy of the interface and the origin of quasi-cleavage cracking.¹³⁾ Furthermore, to suppress quasi-cleavage fracture, a strategy to partition hydrogen inside intermetallic compound (IMC) particles with a higher hydrogen binding energy than that at the MgZn₂ interface has been proposed.¹²⁾ Assuming that the amount of hydrogen in the material does not change due to the dispersion of IMC particles, the increase in hydrogen inside the IMC particles is expected to relatively decrease the hydrogen content at the precipitate interface, thereby suppressing hydrogen-induced quasi-cleavage fracture. The representative IMC particles in practical 7XXX alloys are Al₇Cu₂Fe and Mg₂Si.^14,15) The presence of hydrogen trap sites inside these particles was investigated by Yamaguchi et al. by first-principles calculations.¹⁶⁾ The researchers concluded that Mg₂Si has no internal hydrogen trap sites and that Al₇Cu₂Fe has many hydrogen trap sites in crystal lattices with a maximum hydrogen binding energy of 0.56 eV/atom.¹⁶⁾ In other words, Al₇Cu₂Fe is expected to be a candidate particle for the suppression of hydrogen embrittlement.

The hydrogen embrittlement suppression technique by IMC particles is assumed to trap hydrogen inside the particles, rendering it innocuous and non-contributable to material fracture. However, the hydrogen embrittlement of the particles themselves due to hydrogen trapping (i.e., accelerated damage and decohesion of the particles by hydrogen) is unclear. Even if hydrogen embrittlement can be suppressed by IMC particles, hydrogen-induced damage to the IMC particles leads to premature failure; therefore, the macroscopic mechanical properties of the material are not improved. To establish a novel technique to suppress hydrogen embrittlement with IMC particles, it is imperative to elucidate the influence of hydrogen on the particle damage behavior. In this study, an in situ observation of the tensile deformation process using synchrotron radiation X-ray computed tomography (CT) was performed to quantitatively analyze the damage behavior of the IMC particles in Al–Zn–Mg–Cu alloys, and the influence of hydrogen on the damage behavior of IMC particles was evaluated.

2. Experiment

2.1 Specimens

Al–Zn–Mg–Cu alloys with the chemical compositions shown in Table 1 were used in this study. To evaluate the influence of hydrogen on the damage and decohesion of IMC particles, two types of specimens were prepared, one with high hydrogen (HH) content and the other with low hydrogen (LH) content. After casting, homogenization treatment (743 K, 24 h) and hot rolling (673 K) were performed; only the LH specimen was subjected to two thermal cycles, ranging from room temperature to 633 K to reduce the internal hydrogen content. Subsequently, a solution treatment (748 K, 2 h) was performed, and it was followed by rapid cooling and natural aging (RT, 1 day) and artificial aging (393 K, 6 h; 423 K, 5 h). Figure 1 shows optical micrographs of these plates after mechanical polishing and etching with Keller’s solution. A comparison of Fig. 1(a) and (b) suggests that there is no significant microstructural change due to the thermal cycling described above. Small tensile specimens with gauge lengths of 0.7 mm and cross-sectional areas of 0.6 mm × 0.6 mm were cut from artificially aged plates by electrical discharge machining. From there, the rolling direction of the plate was machined to be the same as that of the tensile direction of the tensile specimen. Details on the specimen geometry are available elsewhere.¹⁷⁾ When cutting the specimens, electrical discharge machining was performed in water for the HH specimen to charge hydrogen and in oil for the LH specimen to suppress the hydrogen uptake. The hydrogen content of the specimens was measured by thermal desorption analysis (TDA). The heating rate was 1.5 K/min, and the temperature ranged from room temperature to 773 K. In this study, a cumulative hydrogen content reaching 773 K was considered to be the internal hydrogen content of the specimen. The hydrogen contents of the LH and HH specimens were 2.4 and 5.4 mass ppm, respectively; these contents were reduced to less than 1/2 of the original values after two thermal cycles. The hydrogen concentration of the Al–Zn–Mg–Cu alloy without electrical discharge machining was reported to be 0.13 mass ppm.¹⁸⁾

Table 1 Chemical composition of the studied alloy (mass%).

Fig. 1

Microstructures of the Al–Zn–Mg–Cu alloy used in this study. (a) and (b) are optical microscope images of the alloy (a) without and (b) with the thermal cycle, respectively.

2.2 High-resolution X-ray CT and in situ tensile test

In situ observation of hydrogen embrittlement behavior using X-ray CT was performed with the SPring-8 BL20XU. Using 20 keV monochromatic X-rays, 1800 projection images were recorded while the specimen on the stage was rotated 180 degrees. In this X-ray CT analysis, a wide-field projection-type X-ray CT with a spatial resolution of 1 µm, which is suitable for a macroscopic observation of the entire specimen, and an imaging-type X-ray CT with a high spatial resolution of 150 nm, which is appropriate for a microscopic observation of the damage evolution, were used for imaging.¹⁹⁾ The tomographic images were reconstructed from the X-ray projection images using the convolution back-projection algorithm. The reconstructed 16-bit cross-sectional image was normalized to 8 bits to ensure that the minimum and maximum values of the linear absorption coefficients correspond to the minimum and maximum values of the gray value. Tomographic in situ tensile tests were performed with a small tensile apparatus on a rotating stage. CT scanning was performed after unloading and just after yielding; the process was repeated at every 0.02-mm displacement until fracture. Under stress, the hydrogen distribution in the material localizes depending on the total deformation time, and hydrogen embrittlement accordingly appears more pronounced.¹²⁾ Due to the limited beam time of synchrotron radiation experiments, the tensile speed in this experiment was set to 0.2 mm/min, and the displacement was held for 55.6 min at each tensile step to facilitate the evaluation of the hydrogen influence. The specimens were scanned in 4 tensile stages for HH and 7 tensile stages for LH specimens before fracture.

2.3 3D/4D image analysis

2.3.1 Quantitative analysis of pores, voids and particles

The quantitative analysis of the volumes, surface areas, and centers of gravity of the pores, voids, and particles were performed with the reconstructed X-ray CT images. These microstructural features were employed as feature points in the local plastic strain calculations described later. The Marching Cubes algorithm was used for calculating the volume and surface area.²⁰⁾ To eliminate fringes and noise existing in the CT images, only objects greater than or equal to 27 voxels were used as feature points.

2.3.2 3D mapping of equivalent plastic strain

Identical microstructural features were tracked throughout all of the tensile stages, and the physical displacement of the features between each tensile stage was measured from the image. First, registration was performed to compensate for the rotation and translation of the material to improve the tracking accuracy of the feature points. This registration was performed for three translational components and three rotational components. Next, identical particles were specified using the matching parameter method^21,22) and tracked at each tensile stage. Delaunay tessellation²³⁾ was used for generating a tetrahedral mesh with the tracked particles as nodes, and the tensile, compressive, and shear strain components were calculated in 3D based on their nodal displacements.

2.3.3 Morphological analysis of particle damage

In this study, particle damage was classified into three types: fracture damaged due to particle fracture, debonded due to decohesion of the interface between the particle and the matrix, and undamaged. Figure 2 shows a schematic illustration of this damage classification algorithm. The volume of each void was expanded by 10 voxels using the image analysis software Amira, and the damage characteristics were classified based on the contact state between the voids and particles. This expansion of voids was aimed at preventing incorrect contact judgments derived from the gradient of gray values at the interface. When a void was in contact with two or more particles (Fig. 2(a)), it was judged as fracture; when it was in contact with a single particle (Fig. 2(b)), it was judged as debonding. A void that was not in contact with a particle (Fig. 2(c)) was classified as undamaged.

Fig. 2

Schematic illustration of the algorithm to classify particle damage morphologies. The damage morphologies of (a), (b), and (c) are assessed as fractured, debonded, and undamaged, respectively.

3. Results and Discussion

3.1 Deformation behavior under hydrogen influence

Figure 3 shows the stress–strain curves measured by the in situ tensile test using synchrotron radiation X-ray CT. The LH specimen fractured at an applied strain of 9.7%, whereas the HH specimen fractured at an applied strain of 5.8%, indicating that the macroscopic elongation was decreased by the presence of hydrogen particles. In terms of stress, the ultimate tensile stress of the HH specimen was lower than that of the LH specimen. This hydrogen-induced decrease in ultimate tensile stress was consistent with the results of similar Al–Zn–Mg–Cu alloys;²⁴⁾ however, this study could not elucidate the reason for the hydrogen-induced stress decrease. The periodic stress drops in the stress–strain curves in Fig. 3 occurred due to stress relaxation from displacement holding during X-ray CT scanning.

Fig. 3

Nominal stress–strain curves obtained from in situ tensile tests in synchrotron X-ray tomography.

Figure 4 shows the equivalent plastic strain distributions in the HH and LH specimens at each tensile stage until just before fracture. In the LH specimen shown in Fig. 4(a)–(e), strain localization appeared at an applied strain of 3.7% and developed with further tensile loading, resulting in the formation of a high-strain region just before fracture. The crack propagated within this high-strain localization region; the fracture path is represented by the black line in the figure. In contrast, the HH specimen in Figs. 4(f)–(h) exhibited crack propagation and fracture along the strain localization region with less strain development than the LH specimen. The local equivalent plastic strain just before fracture (Figs. 4(e) and (h)) reached maximum values of 41.6% for the LH specimen and 20.9% for the HH specimen, indicating that the equivalent plastic strains were localized to values approximately 5 times greater than that of the applied strain. Figures 5(a) and (b) show the equivalent plastic strain maps of LH and HH specimens at the same applied strain stage level, respectively. Figures 5(a) and 5(b) are maps of Figs. 4(a) and 4(g), respectively, where the color scale is changed for comparison. In the LH specimen shown in Fig. 5(a), no strain localization was observed at an applied strain of 2.2%. In the LH specimen shown in Fig. 5(a), no strain localization was observed at an applied strain of 2.2%; in the HH specimen shown in Fig. 5(b), strain was already localized in the vicinity of the fracture path at an applied strain of 2.5%. Although the difference in the applied strain of the two specimens was small (0.3%), the mean equivalent plastic strain was 4.5% for the HH specimen and 3.2% for the LH specimen, resulting in early strain localization in the HH specimen.

Fig. 4

Equivalent strain distributions on the RD–ND cross section of LH and HH specimens. Strain maps of the identical 2D virtual cross section analyzed at ε_a of (a) 2.2%, (b) 3.7%, (c) 6.4%, (d) 7.6%, and (e) 8.8% in the LH specimen and analyzed at ε_a of (f) 1.6%, (g) 2.5%, and (h) 4.2% in the HH specimen are shown.

Fig. 5

3D equivalent strain distribution at similar applied strain steps: (a) LH (ε_a: 2.2%) and (b) HH (ε_a: 2.5%).

3.2 Initiation and growth behavior of voids

The tomographic slices obtained by projection-type X-ray CT at each tensile stage are shown in Fig. 6. Al₇Cu₂Fe, Mg₂Si, and the voids are each shown in different colors, depending on the difference in the linear absorption coefficient of the X-ray from the matrix: white, gray, and black, respectively. In situ observation using synchrotron radiation X-ray CT enabled quantitative analyses of void initiation and growth behavior, as shown in Figs. 6(a)–(f). In the tomographic slices of both specimens, identical particles and voids are surrounded by black dashed lines. The applied strains at void initiation sites were 6.4% and 4.2% for LH and HH specimens, respectively, with no drastic change due to hydrogen. The voids grew with increasing applied strain, but those in the LH specimen were coarser just before fracture, as shown in Fig. 6(d). A comparison of Figs. 6(e) and (i) indicates that the HH specimen exhibited more brittle fracture behavior than that of the LH specimen, where the voids remained small even after fracture. Although the voids were visualized as heterogeneous particle nucleation, it was difficult to determine the void initiation sites, such as particle damage or debonding sites, in the tomographic images of projection-type X-ray CT. Analyses of the void initiation sites, which will be described in Section 3.3, were conducted using tomographic images by imaging X-ray CT.

Fig. 6

Observation of voids initiation and growth using projection-type CT: the tomographic slice of LH specimen captured at ε_a = (a) 2.2%, (b) 3.7%, (c) 6.4%, (d) 7.6%, and (e) after fracture, and the tomographic slice of HH specimen captured at ε_a = (f) 1.6%, (g) 2.5%, (h) 4.2%, and (i) after fracture.

The initiation and growth behaviors of voids were quantitatively analyzed using the matching parameter technique. Figure 7 shows the results of classifying the mean diameters and volume fractions of voids for each initiation strain. Figures 7(a) and (b) show that the growth of pores, which are preexisting voids at a 0% applied strain, was limited in both the HH and LH specimens. This phenomenon suggests that the growth and coalescence of pores are not the dominant driving forces of fracture in these specimens.²⁵⁾ In contrast, the void growth rates were similar regardless of the initiation strain. Figures 7(c) and (d) show the changes in the volume fractions of voids in the HH and LH specimens, respectively. The volume fractions and mean diameters of voids increased with increasing strain, and this trend persisted until fracture. This phenomenon is because the frequency of particle damage increased in the later stages of deformation, resulting in the initiation and coalescence of voids.

Fig. 7

Growth behavior of voids that were initiated at different strain levels. (a) and (b) show the change in size distributions in the HH and LH specimens, respectively. (c) and (d) are the change in volume fractions in HH and LH specimens, respectively.

The early strain localization shown in Fig. 5 and the void initiation and growth behavior shown in Figs. 6 and 7 were combined and evaluated. Figure 8 shows a two-dimensional histogram in which the void-initiated equivalent strain is plotted on the horizontal axis and the void diameter is plotted on the vertical axis. Here, as shown in Fig. 5, the applied strain levels that were analyzed were 2.2% for the LH specimen and 2.5% for the HH specimen, as shown in Figs. 8(a) and (b), respectively. Regarding the plastic strain of void initiation, more voids were initiated in the HH specimen at a higher equivalent plastic strain than that in the LH specimen. This finding indicates that the particles were easily damaged and numerous voids were initiated in the strain-localized region of the HH specimen. In other words, the early particle damage (i.e., void initiation) in this strain-localized region is attributed to the early fracture of the HH specimen. In fact, the total number density of voids was 2.1 × 10¹² m⁻³ and 4.0 × 10¹² m⁻³ for LH (Fig. 8(a)) and HH (Fig. 8(b)), respectively, showing that the HH specimen, at a strain value that was 2.5% the size of the applied strain, had nearly twice the number of voids as the LH specimen.

Fig. 8

Two-dimensional histograms as a function of the size of voids and the local equivalent strain at the voids initiated in (a) LH and (b) HH.

3.3 Damage behavior of Al₇Cu₂Fe and Mg₂Si particles

In this section, we discuss the individual damage behavior of the Al₇Cu₂Fe and Mg₂Si particles. High-resolution tomographic images of the HH specimen at an applied strain of 4.2%, obtained by imaging-type X-ray CT, are shown in Fig. 9. The white and gray particles are Al₇Cu₂Fe and Mg₂Si, respectively. The black regions are voids; the voids in Fig. 9 were formed by the fracture of Mg₂Si and Al₇Cu₂Fe. Figure 9 has a higher resolution than that of the tomographic image of the projection-type X-ray CT shown in Fig. 6, allowing detailed observation and analysis of the particle damage morphology. Figure 10 shows 3D-rendered images of particles and voids using imaging-type X-ray CT. Figures 10(a) and (b) show the 3D images of the HH specimen with an applied strain of 4.2% and the LH specimen with an applied strain of 6.4%, respectively. In terms of particle damage morphology, fracture and debonding were observed in both the HH and LH specimens for both the Al₇Cu₂Fe and Mg₂Si particles. On the other hand, a comparison of the HH and LH specimens shows no clear difference in the particle damage characteristics. Qualitative evaluations of the 3D images show that it is unclear how the presence of hydrogen affected the fracture and debonding of particles.

Fig. 9

Tomographic cross section of imaging-type CT in HH specimen at ε_a of 4.2%.

Fig. 10

3D perspective view of Al₇Cu₂Fe particles, Mg₂Si particles and voids captured at ε_a of 4.2% in the (a) HH specimen and at ε_a of 6.4% in the (b) LH specimen. Al₇Cu₂Fe, Mg₂Si, and voids (pores) are shown in light blue, gray, and red, respectively. The aluminum matrix is not displayed.

Therefore, the fractured, debonded, and undamaged fractions of the particles were quantified statistically and three-dimensionally. Figure 11 shows the results of the damage morphology analysis of Al₇Cu₂Fe. This analysis was performed at the same applied strain levels shown in Fig. 10: 4.2% for the HH specimen and 6.4% for the LH specimen. Figure 11 shows the frequencies of fractured, debonded, and undamaged particles in each bin (each diameter and each sphericity); it uses the number of total particles in each bin as the population. Figures 11(a), (b) and (c) show the frequency of undamaged, fractured and debonded particles, respectively, which were organized by the sphere equivalent diameter; similarly, Figs. 11(d), (e) and (f) show the results organized by sphericity. From Figs. 11(a)–(c), the larger Al₇Cu₂Fe particles tended to cause fracture and debonding in both the HH and LH specimens. Figures 11(d)–(f) show that the particles with lower sphericities were more likely to be damaged than those with higher sphericities in both the HH and LH specimens. In the damaged particles, the frequency of fracture damage was higher than that of debonding, regardless of the hydrogen content.

Fig. 11

Classification analysis of fracture, debonding, and no-damage of Al₇Cu₂Fe particles in HH and LH specimens. The analysis classified (a) no damage, (b) fracture, and (c) debonding as a function of equivalent diameter, and the analysis classified (d) no damage (e) fracture, and (f) debonding as a function of sphericity.

Al₇Cu₂Fe, as described in the introduction, has an internal hydrogen trap site of 0.56 eV/atom,¹⁶⁾ which has better hydrogen trapping abilities than those of dislocations and vacancies; Al₇Cu₂Fe traps hydrogen more effectively than dislocations and vacancies, indicating that Al₇Cu₂Fe has a higher hydrogen concentration than that of the matrix. Although there was concern that the Al₇Cu₂Fe particles themselves might become hydrogen embrittled due to their high internal hydrogen trapping abilities, the results in Fig. 11 clearly show that hydrogen did not enhance particle fracture or debonding. The slight differences in the fracture and debonding behaviors of Al₇Cu₂Fe between LH and HH specimens were not considered to be hydrogen dependent.

In this experiment, hydrogen charging was performed after aging treatment, and hydrogen diffused from the matrix phase into the interior of the Al₇Cu₂Fe particle. The hydrogen diffusion coefficient of Al₇Cu₂Fe was not determined, but it is assumed that the coarser the Al₇Cu₂Fe particle is, the greater the difference in hydrogen content between the interior and surface of the particles. In other words, the finer the Al₇Cu₂Fe particle is, the higher the hydrogen occupancy inside the particle. Therefore, if Al₇Cu₂Fe itself is embrittled by hydrogen, we can estimate that the influence of hydrogen appears in the damage behavior of finer Al₇Cu₂Fe and that the frequencies of fracture and debonding are different between the HH and LH specimens. However, Fig. 11 demonstrates that the difference in the damage behavior of Al₇Cu₂Fe that appeared in the HH and LH specimens was not due to the presence of hydrogen because more fracture and debonding were observed for coarser particles. Differences in mechanical states, such as particle dispersion/agglomeration, local strain distribution near the particles, or plastic constraints around the particles, presumably drive the damage behavior of the Al₇Cu₂Fe particles.

A damage classification analysis was also performed for Mg₂Si, and the results are shown in Fig. 12. Similar to the analysis of Al₇Cu₂Fe, Figs. 12(a), (b) and (c) show the frequency of undamaged, fractured and debonded behaviors arranged by diameter, respectively, and Figs. 12(d), (e) and (f) show the results arranged by sphericity. Figures 12(a)–(c) show that Mg₂Si and Al₇Cu₂Fe were susceptible to damage with large particle sizes. Figures 12(d)–(f) show that Mg₂Si particles with sphericities exceeding 0.6 were more numerous; as with Al₇Cu₂Fe, the particles with lower sphericities tended to be damaged. There were no significant differences between the LH and HH materials in terms of fracture damage, although there was slightly more fracture damage in the LH specimen. Unlike Al₇Cu₂Fe, Mg₂Si had no internal hydrogen trapping sites;¹⁶⁾ therefore, this difference in fracture behavior was not due to hydrogen.

Fig. 12

Classification analysis of fracture, debonding, and no-damage of Mg₂Si particles in HH and LH specimens. The analysis classified (a) no damage, (b) fracture, and (c) debonding as a function of equivalent diameter, and the analysis classified (d) no damage (e) fracture, and (f) debonding as a function of sphericity.

In contrast, in the debonding damage shown in Figs. 12(c) and (f), the Mg₂Si in the HH specimen was more easily debonded than it was in the LH specimen. The Mg₂Si dispersed in the matrix was a few µm in size, and this Mg₂Si interface could be assumed to be incoherent.²⁶⁾ Very recent first-principles calculations have revealed that the Mg₂Si/Al interface is a strong hydrogen trapping site.²⁷⁾ Furthermore, the hydrogen-induced decrease in interfacial cohesive energy reported for the MgZn₂ interface also occurred at the Mg₂Si interface,¹³⁾ resulting in interfacial debonding, as shown in Figs. 12(c) and (f). We concluded that hydrogen influenced the interfacial debonding of Mg₂Si.

The damage behavior of the particles obtained in this experiment will be discussed below. Both Al₇Cu₂Fe and Mg₂Si particles exhibited a trend toward damage with increasing size because coarser particles are prone to having cracks and interfacial voids²⁸⁾ and therefore are more liable to fracture due to these defects. The damage to the less spherical particles can be mechanistically interpreted as the result of local stress concentration at the edges with large curvatures and irregular morphologies.²⁹⁾

For Al₇Cu₂Fe, no hydrogen effect appeared in any of the damage analyses classified by particle size and sphericity. This observation suggests that for IMC particles, such as Al₇Cu₂Fe, that have high hydrogen binding energy inside the particle, hydrogen preferentially accumulates inside the particle rather than at the interface, and thereby hydrogen embrittlement behavior due to interfacial debonding is less apparent. The damage behavior of Al₇Cu₂Fe particles was governed by the mechanical responses of the particles themselves and their vicinity, not by the influences of the hydrogen particles. These results are consistent with the conclusions of a previous report that investigated the influence of hydrogen on the damage behavior of Al₇Cu₂Fe and Mg₂Si particles in an A7075 alloy using projection-type synchrotron radiation X-ray CT.³⁰⁾

No particle hydrogen embrittlement due to internal hydrogen was observed in Al₇Cu₂Fe. However, Al₇Cu₂Fe particles are intrinsically brittle,³¹⁾ and even if dispersed in the matrix to prevent hydrogen embrittlement, abundant voids are formed due to particle damage, resulting in premature fracture. Therefore, to establish hydrogen embrittlement suppression by IMC particle dispersion, it is necessary to find IMC particles that have high hydrogen trapping capacity and high fracture resistance values.

Very recently, high internal hydrogen trapping energies of 0.67 and 0.50 eV/atom were found for Al₁₂Mn₃Si and Al₆Mn, respectively.³²⁾ Because these Mn-based IMC particles could be dispersed in the matrix at a few hundred nm and do not exhibit a distorted shape like Al₇Cu₂Fe,^33,34) they are expected to be less brittle than Al₇Cu₂Fe particles. Further studies are needed to investigate the hydrogen embrittlement suppression behavior of alloys with Mn-based IMC dispersions.

4. Conclusions

In situ observations of hydrogen embrittlement behavior using synchrotron radiation X-ray CT were performed to analyze the influences of hydrogen on the damage behaviors of IMC particles in Al–Zn–Mg–Cu alloys, and the following conclusions were made.

1. (1)    The fracture strains of the HH and LH specimens were 5.8% and 9.7%, respectively, and the HH specimen, which had more internal hydrogen, fractured before the LH specimen.
2. (2)    Hydrogen-induced early-strain-localization during tensile deformation, particle fracture and debonding occurred in these strain-localized regions, resulting in the initiation and growth of numerous voids. In particular, particle damage was severe in the strain-localized region of the HH specimen, which contributed to premature fracture.
3. (3)    Regardless of the hydrogen content, fracture was dominant in the particle damage of Al₇Cu₂Fe, while the damage frequency of fracture and debonding was similar for Mg₂Si particles. Hydrogen had no influence on the particle damage of Al₇Cu₂Fe, but hydrogen enhanced the interfacial debonding of Mg₂Si. The early brittle fracture of Al₇Cu₂Fe depended not on the influence of hydrogen but on the mechanical properties of the particles themselves.

Acknowledgments

This study was partly undertaken with the support of the Grant-in-aid for Scientific Research from JSPS through Subject No. 21K14037, Japan. The financial support from the Light Metal Educational Foundation is also gratefully acknowledged. This work was also supported by JST, CREST Grant Number JPMJCR1995, Japan. The synchrotron experiments were performed at SPring-8 with the approval of the Japan Synchrotron Radiation Research Institute through proposal numbers 2020A1084, 2020A1531, 2021A1002, 2021A1120, 2021B1123, and 2021B1124.

REFERENCES

• 1)   A.  Azarniya,  A.K.  Taheri and  K.K.  Taheri: J. Alloy. Compd. 781 (2019) 945–983. doi:10.1016/j.jallcom.2018.11.286
• 2)   R.  Horiuchi and  H.-S.  Park: J. JILM 24 (1974) 458–466. doi:10.2464/jilm.24.10_458
• 3)   Y.  Shi,  Q.  Pan,  M.  Li,  X.  Huang and  B.  Li: Mater. Sci. Eng. A 621 (2015) 173–181. doi:10.1016/j.msea.2014.10.058
• 4)   Y.  Baba: J. JILM 24 (1974) 227–238. doi:10.2464/jilm.24.227
• 5)   H.K.  Birnbaum and  P.  Sofronis: Mater. Sci. Eng. A 176 (1994) 191–202. doi:10.1016/0921-5093(94)90975-X
• 6)   A.R.  Troiano: Metallogr. Microstruct. Anal. 5 (2016) 557–569. doi:10.1007/s13632-016-0319-4
• 7)   M.  Nagumo: Mater. Sci. Technol. 20 (2004) 940–950. doi:10.1179/026708304225019687
• 8)   M.  Yamaguchi,  M.  Itakura,  T.  Tsuru and  K.I.  Ebihara: Mater. Trans. 62 (2021) 582–589. doi:10.2320/matertrans.MT-M2020375
• 9)   L.  Ismer,  M.S.  Park,  A.  Janotti and  C.G.  Van De Walle: Phys. Rev. B 80 (2009) 184110. doi:10.1103/PhysRevB.80.184110
• 10)   M.  Yamaguchi,  K.I.  Ebihara,  M.  Itakura,  T.  Tsuru,  K.  Matsuda and  H.  Toda: Comput. Mater. Sci. 156 (2019) 368–375. doi:10.1016/j.commatsci.2018.10.015
• 11)   K.  Shimizu,  H.  Toda,  C.  Kadogawa,  H.  Fujihara and  A.  Takeuchi: Materialia 11 (2020) 100667. doi:10.1016/j.mtla.2020.100667
• 12)   H.  Su,  H.  Toda,  K.  Shimizu,  K.  Uesugi,  A.  Takeuchi and  Y.  Watanabe: Acta Mater. 176 (2019) 96–108. doi:10.1016/j.actamat.2019.06.056
• 13)   T.  Tsuru,  K.  Shimizu,  M.  Yamaguchi,  M.  Itakura,  K.  Ebihara,  A.  Bendo,  K.  Matsuda and  H.  Toda: Sci. Rep. 10 (2020) 1998. doi:10.1038/s41598-020-58834-6
• 14)   H.  Su,  T.  Yoshimura,  H.  Toda,  M.S.  Bhuiyan,  K.  Uesugi,  A.  Takeuchi,  N.  Sakaguchi and  Y.  Watanabe: Metall. Mater. Trans. A 47 (2016) 6077–6089. doi:10.1007/s11661-016-3773-3
• 15)   T.  Manaka and  G.  Itoh: J. JILM 67 (2017) 67–71. doi:10.2464/jilm.67.67
• 16)   M.  Yamaguchi,  T.  Tsuru,  K.I.  Ebihara,  M.  Itakura,  K.  Matsuda,  K.  Shimizu and  H.  Toda: Mater. Trans. 61 (2020) 1907–1911. doi:10.2320/matertrans.MT-M2020201
• 17)   M.S.  Bhuiyan,  H.  Toda,  K.  Uesugi,  A.  Takeuchi and  Y.  Watanabe: Mater. Sci. Eng. A 793 (2020) 139423. doi:10.1016/j.msea.2020.139423
• 18)   H.  Su,  H.  Toda,  R.  Masunaga,  K.  Shimizu,  H.  Gao,  K.  Sasaki,  M.S.  Bhuiyan,  K.  Uesugi,  A.  Takeuchi and  Y.  Watanabe: Acta Mater. 159 (2018) 332–343. doi:10.1016/j.actamat.2018.08.024
• 19)   A.  Takeuchi,  K.  Uesugi,  M.  Uesugi,  H.  Toda,  K.  Hirayama,  K.  Shimizu,  K.  Matsuo and  T.  Nakamura: Rev. Sci. Instrum. 92 (2021) 023701. doi:10.1063/5.0020293
• 20)  W.E. Lorensen and H.E. Cline: Proc. 14th Annu. Conf. Comput. Graph. Interact. Tech. SIGGRAPH 1987, Vol. 21, (1987) pp. 163–169.
• 21)   M.  Kobayashi,  H.  Toda,  Y.  Kawai,  T.  Kobayashi,  K.  Uesugi,  D.S.  Wilkinson,  E.  Maire and  Y.  Aoki: J. Japan Inst. Met. Mater. 71 (2007) 181–186. doi:10.2320/jinstmet.71.181
• 22)   M.  Kobayashi,  H.  Toda,  Y.  Kawai,  T.  Ohgaki,  K.  Uesugi,  D.S.  Wilkinson,  T.  Kobayashi,  Y.  Aoki and  M.  Nakazawa: Acta Mater. 56 (2008) 2167–2181. doi:10.1016/j.actamat.2007.12.058
• 23)   C.B.  Barber,  D.P.  Dobkin and  H.  Huhdanpaa: ACM Trans. Math. Softw. 22 (1996) 469–483. doi:10.1145/235815.235821
• 24)   M.S.  Bhuiyan,  Y.  Tada,  H.  Toda,  S.  Hang,  K.  Uesugi,  A.  Takeuchi,  N.  Sakaguchi and  Y.  Watanabe: Int. J. Fract. 200 (2016) 13–29. doi:10.1007/s10704-016-0092-z
• 25)   H.  Toda,  H.  Oogo,  K.  Horikawa,  K.  Uesugi,  A.  Takeuchi,  Y.  Suzuki,  M.  Nakazawa,  Y.  Aoki and  M.  Kobayashi: Metall. Mater. Trans. A 45 (2014) 765–776. doi:10.1007/s11661-013-2013-3
• 26)   F.  Kurosawa and  T.  Akimoto: J. Japan Inst. Met. Mater. 71 (2007) 641–645. doi:10.2320/jinstmet.71.641
• 27)   M.  Yamaguchi et al.: unpublished work.
• 28)   J.J.  Williams,  Z.  Flom,  A.A.  Amell,  N.  Chawla,  X.  Xiao and  F.  De Carlo: Acta Mater. 58 (2010) 6194–6205. doi:10.1016/j.actamat.2010.07.039
• 29)   V.  Tvergaard: J. Mechan. Phys. Solids 41 (1993) 1309–1326. doi:10.1016/0022-5096(93)90081-P
• 30)   Y.  Wang,  H.  Toda,  Y.  Xu,  K.  Shimizu,  K.  Hirayama,  H.  Fujihara,  A.  Takeuchi and  M.  Uesugi: Acta Mater. 227 (2022) 117658. doi:10.1016/j.actamat.2022.117658
• 31)   S.S.  Singh,  E.  Guo,  H.  Xie and  N.  Chawla: Intermetallics 62 (2015) 69–75. doi:10.1016/j.intermet.2015.03.008
• 32)   K.  Shimizu et al.: unpublished work.
• 33)   H.  Nakayasu,  E.  Kobayashi,  T.  Sato,  R.  Holmestad and  K.  Marthinsen: Mater. Charact. 136 (2018) 367–374. doi:10.1016/j.matchar.2017.12.006
• 34)   A.M.F.  Muggerud,  Y.  Li and  R.  Holmestad: Philos. Mag. 94 (2014) 556–568. doi:10.1080/14786435.2013.857796